Directory UMM :Data Elmu:jurnal:I:International Review of Law and Economics:Vol19.Issue4.Dec1999:

Discrimination and Sexual Harassment

RICHARDA. POSNER

University of Chicago Law School, Chicago, Illinois, U.S.A.

This is the era of deregulation—and yet in U.S. labor markets, at least, the legal regulation of the employment relation has been expanding in recent decades. The laws have been stiffened and their enforcement, mainly through private lawsuits, has been beefed up. Currently the two most important areas, in terms of impact on employers, are age discrimination and sexual harassment, and these will be my focus. © 2000 by Elsevier Science Inc.

I. Age Discrimination

The Nature and Consequences of Age Discrimination in Employment

Animus discrimination. The justification that is most commonly offered for forbidding age discrimination is that people over 40 are subject to a form of prejudice, “ageism,” that is analogous to racism and sexism. It comes in two forms. One, called “animus discrimination,” is a systematic undervaluation, motivated by ignorance, viciousness, or irrationality, of the value of older people in the workplace. There is undoubtedly resentment of and disdain for older people, especially in American society, and wide-spread misunderstandings about aging and old age, some of which are disadvantageous to the old.1 _{For example, middle-aged and elderly people may seem old-fashioned} because they “cling” to “outmoded” methods, yet the outmoded methods may be “clung to” only because theincrementalbenefit of the latest method is slight, so that anyone who is invested in the old method is rationally reluctant to switch.

This is the revised text of an address given on September 24, 1998, at the Annual Meeting of the European Law and Economics Association. The article draws on previous writings of mine:Aging and Old Age, ch. 13 (Chicago: University of Chicago Press 1995); and “Status Signaling and the Law, with Particular Application to Sexual Harassment,” 147 University of Pennsylvania Law Review1069 (1999), coauthored with Gertrud M. Fremling. I thank Susan Burgess for helpful research assistance.

1_{Some, however, are advantageous. There is a tendency to ascribe the maturity, wisdom, and disinterest possessed}

by some old people to all or most of them. Elissa T. Perry, Carol T. Kulik, and Anne C. Bourhis, “Moderating Effects of Personal and Contextual Factors in Age Discrimination,” 81Journal of Applied Psychology628, 643 (1996). Also, because few people understand selection bias, they tend to generalize from the impressive performance of octogenar-ian judges that octogenaroctogenar-ians have unsuspected capabilities, not realizing that the advanced age at which most judges are appointed operates to draw judges from an unrepresentative slice of the aging population.

International Review of Law and Economics 19:421– 446, 1999

© 2000 by Elsevier Science Inc. 0144-8188/99/$–see front matter

But competitive pressures for rational behavior are considerable in private markets and inimical to irrational discrimination,2_{and, besides that, both the people who make} employment policies for corporate and other employers and most of those who carry out those policies by making decisions about hiring or firing specific workers are themselves middle-aged rather than young. They are unlikely to harbor misunderstand-ings about the abilities of workers in this age group3_{(and as a practical matter it is the} middle-aged who complain about age discrimination—most elderly people are volun-tarily retired). Although such workers do have trouble finding new jobs at high wages, this is because the wages in their old jobs will have reflected firm-specific human capital that disappeared when they left and that they cannot readily replace because of the cost of learning new skills and because the proximity of these workers to (voluntary) retirement reduces the expected return from investing in learning new skills.4 _And empirical findings that workers 65 years of age or older perform their jobs as well as younger workers in the same enterprise are vitiated by selection bias: Any demonstrably unsatisfactory older workers will have been fired or nudged into retirement.

Statistical discrimination. The second and more plausible form of ageism (if it should be called that) consists of attributing to all people of a particular age the characteristics of the average person of that age. This is an example of “statistical discrimination,” that is, the failure or refusal, normally motivated by the costs of information, to distinguish a particular member of a group from the average member. Age, like sex, is one of the first facts that we notice about a person and use to “place” him or her. We do this because we operate with a strong, though often an unconscious, presumption, echoing the rigid age grading that structures activities and occupations in many primitive societies, that particular attitudes, behaviors, and positions in life go with particular ages.5 _{Although the presumption that age matters in these ways is rational, there is a} great deal of variance in the capacities, behaviors, and attitudes of persons in particular age groups and, partly as a result, great overlap between the capacities of persons in different age groups. People age at different rates and from different levels of capacity. So if age is used as a proxy for attributes desired or disliked by an employer, some people who are entirely competent to perform to the employer’s specifications will not

2_{For an excellent discussion, see James J. Heckman, “Detecting Discrimination,” 12Journal of Economic Perspectives,}

Spring 1998, p. 101.

3_{Even in a study in which students, not employers, were tested for age bias, very little was found. See Perry et al., note}

1 above, at 641. And the study contains no evidence that such bias as it found was irrational. One study, which found that “nearly 90 percent of [elderly] job losers’ wage reductions are explained by the nontransferability of the workers’ firm-specific skills and knowledge or seniority,” ascribed the remaining 10% to age discrimination, but they had no basis for this ascription. The 10%, as they acknowledged, was merely “a residual remaining after accounting for other factors”—and among the factors not accounted for was a possible age-related decline in capability. As the study was of the wages in new jobs of elderly workers who had lost their previous jobs, the possibility that the workers sampled were underperformers was indeed a significant one. David Shapiro and Steven H. Sandell, “The Reduced Pay of Older Job Losers: Age Discrimination and Other Explanations,” inThe Problem Isn’t Age: Work and Older Americans37, 45, 47 (Steven H. Sandell ed., New York: Praeger 1987). The very next essay in the collection, a study of young and old workers employed by the same firm, finds that theentiredifference in wages between the two groups is due to differences in investment in human capital.

4_{Dian E. Herz and Philip L. Rones, “Institutional Barriers to Employment of Older Workers,” 112Monthly Labor}

Review, April 1989, pp. 14, 20.

5_{Arlie Russell Hochschild,The Unexpected Community21 (Englewood Cliffs, N.J.: Prentice-Hall 1973). For a brief but}

serviceable summary of theories of statistical discrimination, see Paula England, “Neoclassical Economists’ Theories of Discrimination,” inEqual Employment Opportunity: Labor Market Discrimination and Public Policy, 59, 60 – 63 (Paul Burstein ed., New York: Aldine de Gruyter 1994).

be hired, or will be fired or forced to retire to make way for young people who actually are less able—though some elderly people may actually benefit from a presumption that “age matters.”6

This does not, however, make age discrimination in employment inefficient any more than the substitution in some other field of activity of a rule (for example, do not drive faster than 65 miles per hour) for a standard (do not drive too fast for conditions) need be inefficient. A rule is simpler to administer than a standard and, therefore, cheaper, and the cost savings may exceed the loss from disregarding circumstances that may make the rule disserve the purposes behind it in a particular case. Rules have higher error costs but lower administrative costs; standards have lower error costs but higher administrative costs. The relative size of the two types of cost will determine the efficient choice between the alternative methods of regulation in particular settings. Statistical discrimination is an example of rule-based behavior, and because it is a method of economizing on information costs, we can expect it to be more common in settings where those costs are high. Few of us would be comfortable if airline pilots or military officers could not be forced to retire at any age without proof of individual unfitness. Mandatory retirement—a blanket rule against retaining a worker who has reached a specified age, regardless of the particular worker’s actual productivity— has three sup-ports besides the general benefits of a rule. First, knowing far in advance the age at which one will retire facilitates an individual’s financial and retirement planning. A person could always decide that he was going to retire at some particular age, yet he might fear that he might change his mind. Second, because full social security benefits are available at age 65 and are sharply reduced until 70 if the recipient continues working after reaching 65, there are powerful financial advantages to retiring at 65. If, therefore, few workers would want to continue working after that point, the benefits from individualized assessment of their fitness to do so will be small, yet there are apt to be significant fixed costs of establishing and operating the requisite machinery of assessment.

Third, if, as is plausible, a significant decline in a worker’s performance is probable within a few years after he reaches 65, the benefits from individualized assessment will be reduced further because they will be realized for only a short period. The costs of such assessment will rise, moreover, because the employer will have to monitor the performance of workers who have reached the stage of life at which a decline in job performance is highly probable more carefully than the performance of younger workers.

If employers are forbidden to use efficient methods of evaluation, their labor costs will rise, and it is now generally accepted that increases in payroll taxes or other labor costs are borne largely by the workers themselves, in the form of reduced wages or benefits.7 _{The increase in cost operates as a tax, and the incidence of a tax does not} depend on which side of the market (here, employer, or employee) the tax is assessed on.8_{Workers as a whole, few of whom either are wealthy or are guilty of “ageism,” will} in effect be taxed for the benefit of elderly workers, though the elderly—who today are

6_{See note 1 above.}

7_{See Jonathan Gruber and Alan B. Krueger, “The Incidence of Mandated Employer-Provided Insurance: Lessons}

from Workers’ Compensation Insurance,” 5Tax Policy and the Economy111 (1991), and studies cited in id. at 117–118.

8_{See, for example, Joseph A. Pechman,Federal Tax Policy223–224 (5th ed., Wash., DC: Brookings Institution 1987);}

Laurence Kotlikoff and Lawrence Summers, “Tax Incidence,” in Handbook of Public Economics1043, 1047 (Alan Auerbach and Martin Feldstein eds., Amsterdam: North-Holland 1987).

prosperous recipients of substantial public largesse—no longer constitute an oppressed class.

Anyway, the law cannot, merely by outlawing particular types of discrimination, force employers to judge every worker as an individual. The costs of information are too high. Variability in performance in an age cohort tends to grow as the cohort ages. The greater the variability in a population of workers, the longer the employer will have to search to find workers suitable to his needs, unless he relies on some simple proxy or rule of thumb. The costs of individualized assessment will be high. Increased variability also may make the value of such assessments greater— but remember that the expected return from selecting the very best worker (if he is old) will be truncated because an older worker is unlikely to continue working for long.

Deprived of the age proxy, some employers will use other proxies for ability or performance, such as test results, thus “discriminating” against workers whose perfor-mance those proxies underpredict. Airlines, for example, if forbidden to impose man-datory retirement on their pilots, might raise their standards of physical fitness, with the result that some perfectly competent young pilots might be forced out. Some employers denied the use of the age proxy will throw up their hands and, unable to distinguish between good and bad workers of the same age, treat both groups indiscriminately, with the result that bad workers will benefit at the expense of good ones. This is just another form of statistical discrimination. The victims of the two different forms of statistical discrimination—the victims of lumping together people of disparate abilities though the same age (the form of statistical discrimination that the law encourages), or of treating separately people who have the same abilities but are of different age (the form of statistical discrimination that the law forbids)—will be different. But there will still be victims. And the victims of the age discrimination law, as distinct from the victims of what the law calls age discrimination, may be people more marginal, more necessitous, than the average elderly worker. Some evidence for this conjecture is that, as we shall see, most plaintiffs in age discrimination cases are not “workers” at all, but managers, professionals, and executives.

It is true that, as with other forms of discrimination, statistical discrimination against the aged worker may impose an external cost. If the exceptional aged—those young in mind, body, and spirit— cannot cash in on their exceptionality in the employment market because very few employers will look behind chronological age in making employment decisions, they will have a suboptimal incentive to invest in their human capital because the payback period will be artificially truncated. The resulting under-investment in human capital is the joint product of individual decisions by a multitude of employers no one of whom would be better off incurring substantial costs to identify the handful of exceptional elderly workers. It is not a complete answer that an individ-ual worker might by accepting a lower wage induce the skeptical employer to make an exception in his favor. The prospect of having to offer such a wage cut would reduce the expected return to the exceptional employee’s investment in his human capital and, thus, the amount that he would be willing to invest. Also, as I shall point out, the age discrimination law may as a practical matter rule out such transactions.

But because prohibiting the use of the age proxy just leads to the substitution of other proxies, the problem of underinvestment in human capital is shifted, not solved, by the age discrimination law. Whoever is “unfairly” disadvantaged by the new proxy, in the sense that it does not measurehisabilities accurately (perhaps he does not do well on pen-and-pencil tests because of deficiencies in his formal education, but is an excellent worker nevertheless), will lack the incentive to make the optimal investment in his

human capital. And if the new proxies are less efficient than age—as is likely, because otherwise they would probably have been adopted without government prodding— wages will fall because employers’ labor costs will be higher. And with lower wages, there will be less incentive for workers to invest in their human capital.

B. The Effects of the Age Discrimination in Employment Act

Let me consider a bit more systematically the probable effects of the age discrimination law on elderly workers and on the rest of society. The first thing to note is the misfit between the scope of the law and the concerns of the elderly. The law kicks in when a worker turns 40 years of age, and only 10% of the plaintiffs in a recent sample of cases (about which more in a moment), including those plaintiffs who challenged mandatory retirement, are 65 or older—a smaller percentage than the percentage of elderly people in the U.S. population as a whole. The main reason is plain enough: Most people who are 65 or older are voluntarily retired, and so are not protected by the Age Discrimi-nation in Employment Act. Yet the Act was “sold” by means of emotional rhetoric concerning the plight of the elderly, in 1967 still viewed as a disadvantaged segment of American society.

Hiring cases. The cost of training an older worker is higher than that of training a younger one because of the age-related decline of what cognitive psychologists aptly call “fluid intelligence,” whereas the expected return to the investment in training is lower because the older worker has a shorter working life expectancy.9_{The age discrimination} law adds to the costs of employing older workers, and hence to the reluctance of employers to employ them, by giving them more legal rights against their employer than younger workers have. It is true that the Act forbids age discrimination in hiring as well as in firing, demotions, wages, and so forth. But it is largely ineffective against hiring discrimination because of the extreme difficulty of proving substantial damages in such cases. Damages are not the only relief available in a suit under the Act; injunctive relief is also possible. But the disappointed applicant is unlikely to be satisfied with an order requiring the employer to hire him; he would be entering into the employment under most inauspicious circumstances.

Substantial damages are difficult to prove in hiring cases because the plaintiff-applicant, if hired, would probably have received a wage only slightly higher than his reservation wage. For if the job he applied for pays much more than his present job, he will have great difficulty persuading a jury that he was the best-qualified applicant.10_The monetary stakes in a discharge case will often be much greater. If the discharged employee’s wage contained a return for firm-specific human capital, that wage will be higher, maybe much higher, than he could get elsewhere— especially if he is too close to retirement age or too inflexible with regard to learning new skills for a new employer to think it worthwhile to invest in creating new specific capital for him. The difference between what the old worker was paid before he was fired and the much lower wage that

9_{For empirical evidence of employers’ reluctance to hire older workers, see Robert M. Hutchens, “Do Job}

Opportunities Decline with Age?” 42Industrial and Labor Relations Review89 (1988). Although health costs are also higher for older workers, it is not a violation of the age discrimination law for the employer to take these costs into account in designing a wage-benefits package for his employees.

10_{If his next best wage was $20,000, it is hardly likely that, if only he had been younger, the defendant would have}

is the best he can hope for in a new job provides the measure of his compensatory damages.

So it is no surprise that a large sample of litigated age discrimination cases contained

nohiring cases; more than two thirds of the cases involved termination (discharge or involuntary retirement), with most of the others involving promotion or demotion.11_A more recent study of age discrimination complaints lodged with the Equal Employment Opportunity Commission (EEOC) finds that 87.9% of the complaints in which no other form of discrimination was alleged besides age discrimination involved termination, and only 8.6% involved hiring.12_{In a study that I conducted of all court cases under the} Age Discrimination in Employment Act in which a final decision was rendered in 1993 on other than procedural grounds and was reported in Westlaw, the West Publishing Company’s computerized database of judicial decisions, I too found that hiring cases are relatively rare— only 12.3% of the total cases in my sample. And plaintiffs won only one of the hiring cases—a winning percentage of only 2.8%. In general, plaintiffs did very poorly, winning only 11.6% of the cases they brought.13_{Such success as plaintiffs} did have was largely confined to cases that involved a termination. No plaintiff com-plaining about a promotion or a demotion obtained a money judgment.

A low winning percentage for plaintiffs in a class of cases is not (provided it exceeds zero!) conclusive evidence that these cases are “losers” for plaintiffs to bring. It could be an effect of high damage awards. The higher the award if the plaintiff wins, the likelier he is to sue even if the probability of winning is small. It is like a lottery: The bigger the pot, the longer the odds that the organizers of the lottery can set and still sell tickets. And a winning plaintiff in an age discrimination suit is entitled to reimbursement of his attorney’s fees by the defendant, on top of any damages awarded, whereas a losing plaintiff can be ordered to pay the defendant’s attorney’s fees only if the suit was frivolous.

But in the 21 cases in which the plaintiff obtained damages in lieu of or in addition to equitable relief, the average damage award was $184,588. This is an unimpressive figure when one considers not only that the risk of winning nothing is very great—so that when averaged together with the cases in my sample in which the defendant won, the total damages awarded come to only $13,894 per case, a very small expected gain for a federal case litigated all the way to final judgment— but also that cases involving large stakes are likely to be overrepresented in a sample of cases litigated to judgment. For, other things being equal, the greater the stakes in a case, the more likely the case is to be litigated rather than to settle.14

If courts awarded a winning plaintiff attorney’s fees large enough to compensate the plaintiff’s lawyer,ex ante, no matter how small the probability of his winning, then unless lawyers were risk averse and could not assemble a large enough pool of cases to

11_{Michael Schuster and Christopher S. Miller, “An Empirical Assessment of the Age Discrimination in Employment}

Act,” 38Industrial and Labor Relations Review64, 71 (1984) (Table 3); see also Michael Schuster, Joan A. Kaspin, and Christopher S. Miller, “The Age Discrimination in Employment Act: An Evaluation of Federal and State Enforcement, Employer Compliance and Employee Characteristics: A Final Report to the NRTA-AARP Andrus Foundation” iv (unpublished, School of Management, Syracuse University, June 30, 1987).

12_{George Rutherglen, “From Race to Age: The Expanding Scope of Employment Discrimination Law,” 24Journal}

of Legal Studies491, 508 (1995) (Table 5).

13_{While this was 45.9% of the 20.9% of the cases that went to verdict, 69.9% of the cases were disposed of by}

summary judgment, meaning that there was no triable issue; and plaintiffs won only 1.5 percent of those cases. The sum of 20.9% and 69.9% is only 90.8%, the remaining cases were disposed of by other forms of judgment.

eliminate the risk of losing a particular case in the pool, every case in which the probability of the plaintiff’s winning exceeded zero would be brought. But courts usually award the winning plaintiff the attorney’s fee he actually incurred, with no multiplier to reflect the risk of loss. In the seven cases in my sample in which the award of attorney’s fees is disclosed, the average award ($116,093) was 62.9% of the damage award. Projected to the entire sample, this would raise the expected judgment (damages plus attorney’s fees) from $13,984 to $22,633, which is still a very modest amount for a federal case litigated to judgment.

When one considers that more than 100 million people are employed in the United States, a large percentage of them over 40 years of age, the total number of suits that left a trace in Westlaw, and even the 20,000 complaints lodged with the EEOC, are meager. The outcomes of the cases in my sample suggest why.

Damages in a hiring case are likely to be even lower on average than in a termination case. (The award in the sole hiring case in my sample that the plaintiff won was only $63,000.) My guess is that such cases are brought mainly by inexperienced lawyers. Experienced lawyers rarely litigate federal cases in which both the probability of win-ningandthe judgment if the case is won are low; the expected cost of the suit is likely to swamp the expected benefit. The causality could run in the opposite direction; hiring plaintiffs losebecausethey are represented by inexperienced lawyers. But this is unlikely. If experienced lawyers can win a particular type of case, they will be drawn to it.

Firing (and other discharge) cases. The merits of awarding substantial damages even in cases of termination are doubtful from an economic standpoint. Part of a wage that reflects the greater value to the firm of an employee who has firm-specific human capital may be generated by the employer’s contribution to that capital. The employee himself pays for the employer’s investment in the employee’sgeneralhuman capital— skills that he can employ elsewhere in the economy— by accepting a lower wage. The employer would have no protection against the employee’s using that capital to obtain a higher wage from another employer, and he therefore will not pay for it. But the employer will pay for a share, possibly a very large share, of the employee’s firm-specific capital, because the employee by definition cannot obtain a return on that capital from another employer. The greater the share of this capital that the employer pays for, the lower will be the turnover of employees (for they will have a higher salary), and, hence, the likelier will the employer be to reap the benefit of its investment.15

This analysis argues for allowing the employer to deduct from damages in an age discrimination firing case that portion of the difference between the wage he paid the employee, and the lower wage that the employee would command in his best alternative employment, that represents a return on specific human capital for which the employer paid. The employer loses this investment in the employee’s human capital by losing the employee, and should not have to pay twice—first by swallowing the loss of the invest-ment, and then by in effect “buying” the investment back by having to pay damages measured by its value. This point shows, moreover, that employers have their own incentives, unrelated to law, to avoid firing competent employees of any age, even if replacements are available. The employer has invested in the employee, and if the employee is still productive, the employer is continuing to earn a return on the

15_{Gary S. Becker,Human Capital: A Theoretical and Empirical Analysis, with Specific Reference to Education, 33– 49 (3d ed.,}

investment. But it is a point primarily aboutprivateemployers, and the age discrimina-tion law also applies to public employment and to employment by colleges, universities, foundations, and other not-for-profit employers. Public and not-for-profit employers can be expected to discriminate more than private for-profit employers. They face fewer market pressures to minimize their labor costs, and the constraint on their obtaining profits gives them an incentive to substitute nonpecuniary for pecuniary income. One form of nonpecuniary income is avoiding undesired personal associations.16_{Of the 256} cases in my sample for which the necessary information is available, 20.2% were brought against government employers and 6.9% against nonprofit employers. These percent-ages greatly exceed the percentpercent-ages of the labor force for which these two classes of employer account.

Why do age discrimination plaintiffs do poorly? When the Age Employment in Discrim-ination Act was enacted back in 1967, many employers were practicing age discrimina-tion (primarily of the statistical sort) and were doing so openly. It took some time for the message that age discrimination was now an unlawful practice, which, if continued, must be concealed to filter down to the corporate personnel who make the actual employment decisions. They continued for some years blithely to generate “smoking gun” evidence of age discrimination. By now, however, employers have largely suc-ceeded in purging such slogans as “you can’t teach an old dog new tricks”17_{from the} vocabulary of their supervisory and personnel staffs.18_{Some evidence that age} discrim-ination cases are indeed increasingly difficult for plaintiffs to win comes from compar-ing the winncompar-ing percentage of plaintiffs in my sample (less than 12%) with the much higher percentage in the Schuster sample (32%), which was drawn from cases decided between 1968 (the first year after the enactment of the age discrimination law) and 1986.19

In the absence of smoking-gun evidence of age discrimination, now difficult to come by, a plaintiff must as a practical matter show that an equally competent but younger employee was treated better. Such proof is difficult because of the intangible elements in evaluating a worker’s performance other than in the simplest jobs—and the simplest jobs do not generate the plausibly high damage claims that repay the costs and uncertainties of litigation. The simplest jobs require little human capital, whether general or specific, and (partly for that reason) pay low wages.

Moreover, a firm that wants to get rid of an older employee can often do so with near impunity by cashiering a younger employee at the same time. One hears rumors that

16_{See Armen A. Alchian and Reuben A. Kessel, “Competition, Monopoly, and the Pursuit of Money,” inAspects of}

Labor Economics157 (National Bureau of Economic Research 1962).

17_{Or “old dogs won’t hunt,” held evidence of age discrimination inSiegel v. Alpha Wire Corp., 894 F.2d 50, 55 (3d Cir.}

1990).

18_{“It is important tosensitize all managers to the fact that any type of age reference, even in informal conversations, may have}

a negative impact on the organization’s position[in an age discrimination suit].” Robert A. Snyder and Billie Brandon, “Riding the Third Wave: Staying on Top of ADEA Complaints,” 23Personnel Administrator, Feb. 1983, pp. 41, 45 (emphasis in original).

19_{See references in note 11 above. It is interesting to note that the 12% win rate of the plaintiffs in my sample is only}

a little more than half the win rate of plaintiffs in all employment discrimination cases (that is, including racial, ethnic, and sexual discrimination cases along with age discrimination cases). See John J. Donohue III and Peter Siegelman, “Law and Macroeconomics: Employment Discrimination Litigation over the Business Cycle,” 66Southern California Law Review709, 756 (1993) (Figure 4); Theodore Eisenberg, “Litigation Models and Trial Outcomes in Civil Rights and Prisoner Cases,” 77Georgetown Law Journal1567, 1578 (1989). But these studies, like the Schuster studies of age discrimination, deal with an earlier period (1977 to 1988) than my study, and so are not strictly comparable.

this is a common practice. There is high turnover among young employees anyway, and the firm may not yet have invested much in the young employee’s firm-specific human capital (a principal reasonwhyturnover of young employees is high) and so has little to lose from firing him, though concern with reputation must inhibit this Machiavellian strategy to some and perhaps to a great extent.20_{The “RIF” (reduction in force) is a} related strategy21_{; victims of RIFs who complain on age discrimination grounds do very} poorly—in my sample, even more poorly than victims of alleged age discrimination in hiring.

Another thing that makes it hard for the employee to win an age discrimination case is that older employees tend to be more costly to a firm than younger ones, by virtue of receiving a larger package of wages and benefits. The more costly they are, the more difficult it is to ascribe their discharge to their age, as distinct from their expense. The older employee may be more productive by reason of his greater experience, or he may be paid a higher wage either to discourage shirking in his last period of employment or as a reward for not having shirked previously. If he is more productive, then he is not in fact more costly to the firm than a younger, less well paid, but also less productive worker. And if he is being paid a so-called “efficiency” wage either to discourage shirking or to repay the “bond” that he posted as a young employee by accepting a lower salary in exchange for an implicit promise of compensation later if he behaved, he is merely receiving the benefit of his bargain. But a court is not apt to tumble to the reason why the older employer is not really being “overpaid” and to see therefore that the employer is reneging on an implicit contract. All the court can see is that the employer had a reason unrelated to age for firing the older worker— he was more expensive.22 This analysis suggests that age as such is unlikely to be a good predictor of the likelihood of the plaintiff’s winning an age discrimination suit. The older the employee, the easier it will be for the employer to make a plausible case that the employee was fired because he was failing or too expensive, and not because of his age as such. The length of time the plaintiff was employed by the defendant is likely to be a better predictor of the likelihood of the plaintiff’s winning. It is a proxy for the amount of specific human capital invested in him, hence the amount of his damages, hence the likelihood of his having sufficiently large expected damages to be able to attract a competent lawyer to represent him. It is true that the larger the expected gain from suit to the plaintiff, the larger the expected loss to the defendant, who therefore can be expected to defend more vigorously the higher the stakes. But there is a double asymmetry. First, there is some threshold of expected gains from suit below which a potential plaintiff cannot make a credible threat to sue. Second, the defendant normally has more to lose in an age discrimination case than the plaintiff has to win, because a

20_{Cf. Christopher Cornwell, Stuart Dorsey, and Nasser Mehrzad, “Opportunistic Behavior by Firms in Implicit}

Pension Contracts,” 22Journal of Human Resources704 (1991).

21_{See, for a comprehensive discussion that reaches conclusions similar to mine, Gary Minda, “Opportunistic}

Downsizing of Aging Workers: The 1990s Version of Age and Pension Discrimination in Employment,” 48Hastings Law Journal511 (1997).

22_{Courts could try to deal with this problem by treating “discrimination” based on salary as a form of age}

discrimination, because age and salary tend to be positively correlated. But maybe because this would make it difficult for firms to take rational steps to reduce their costs when they find, for whatever reason, that they are paying wages in excess of the market, the Supreme Court seems to have rejected this approach.Hazen Paper Co. v. Biggins, 508 U.S. 948 (1993). That at least is the interpretation ofHazenadopted in such cases asAnderson v. Baxter Healthcare Corp., 13 F.3d 1120, 1125–26 (7th Cir. 1994), andHamilton v. Grocers Supply Co., 986 F.2d 97 (5th Cir. 1993).

victory for the plaintiff will encourage other suits against the defendant.23_This asym-metry is greater, the smaller the monetary stakes, because those stakes are symmetrical. So it will be harder for plaintiffs to win small cases than big ones. In a small case, the plaintiff will find it impossible to hire an excellent lawyer because the expected gains from suit are so slight, whereas the defendant will be willing to pay to hire such a lawyer because the firm will fear the effect of losing the suit on the number of future claims against it.

This analysis may explain why, in my sample, length of service, which as I have explained is a proxy for the amount of specific human capital and hence for the size of the stakes in the case, has a positive correlation with the probability of the plaintiff’s winning but age has—ironically—a negative correlation. Age may be a proxy for not being able to perform to the employer’s legitimate (or at least unprejudiced) expecta-tions. And because expected damages are higher in firing than in other age discrimi-nation cases, firing cases are more likely to be brought by plaintiffs whose probability of winning is slight.

Early retirement offers. At least one important aspect of the Age Discrimination in Employment Act has, it might seem, surely been effective without much litigation: the prohibition of mandatory retirement at fixed ages. The existence of a policy of man-datory retirement is not concealable, so it is doubtful that much litigation has been required to extirpate the practice. Yet employers can, without violating the law and often without incurring heavy other costs, manipulate the age distribution of their employees through offers of early-retirement benefits.24_{The right to take early} retire-ment on terms sufficiently advantageous to make the exercise of the right attractive is a form of employee compensation no different from any other fringe benefit. Its value is uncertain but so is that of health or life insurance, the value of which to an individual worker depends on his individual health experience and longevity, which cannot be known with anything approaching certainty in advance. The more munificent the early-retirement offer, therefore, the less the employer need pay in wages and other benefits.

The creation of a generous early-retirement program funded (in equilibrium) by a reduction in the wage level would, it is true, make the employer more attractive to workers who set a high value on leisure relative to pecuniary income,25 _{whereas the} employer might prefer workers with a stronger work ethic. The more employers adopt

23_{As noted in Rutherglen, note 12 above, at 514.}

24_{Michael C. Harper, “Age-Based Exit Incentives, Coercion, and the Prospective Waiver of ADEA Rights: The Failure}

of the Older Workers Benefit Protection Act,” 79Virginia Law Review1271, 1278 –1279 (1993). For evidence that such offers are effective in bringing down the average age of the workforce, see Laurence J. Kotlikoff and David A. Wise,The Wage Carrot and the Pension Stick: Retirement Benefits and Labor Force Participation(Kalamazoo, MI: W.E. Upjohn Institute for Employment Research 1989); James H. Stock and David A. Wise, “Pensions, the Option Value of Work, and Retirement,” 58Econometrica1151 (1990); Rebecca A. Luzadis and Olivia S. Mitchell, “Explaining Pension Dynamics,” 25Journal of Human Resources679 (1991); and Robert M. Lumsdaine, James H. Stock, and David A. Wise, “Pension Plan Provisions and Retirement: Men and Women, Medicare, and Models,” inStudies in the Economics of Aging183 (David A. Wise ed., Chicago: University of Chicago Press 1994). And for specific evidence that employers use early-retirement offers to “get around” legal restrictions on mandatory retirement, see Edward P. Lazear, “Pensions as Severance Pay,” inFinancial Aspects of the United States Pension System57, 82– 84 (Zvi Bodie and John B. Shoven eds., Chicago: University of Chicago Press 1983).

25_{For empirical evidence, see Olivia S. Mitchell and Gary S. Fields, “The Economics of Retirement Behavior,” 2}

such programs, however, the less will be the effect on the composition of any given employer’s workforce.

Another factor holding down the net cost to the employer of offering early retire-ment is that the risk to an employee of turning down even a rather chintzy such offer may be so great that offers of early retirement need not be princely to induce wide-spread acceptance. Unless the employee can prove that the employer’s package of retirement and other benefits is not abona fidebenefits plan, but is instead designed to evade the statute’s prohibition against age discrimination—and that is not an easy thing to prove—the employer can penalize the employee for refusing an offer of early retirement by offering lower benefits to employees who retire later.26_{The making of an} offer of early retirement, moreover, does not commit the employer to retaining until normal retirement age an employee who turns down the offer. Herein lies the greatest risk to the employee of turning the offer down. An offer of early retirement usually reflects a desire by the employer to reduce the number or average age of its employees, and if not many employees take up the offer, and even if many do, the employer may resort to other measures for achieving the desired size and composition of his work force. The employee knows this and knows, therefore, that if he turns down an offer of early retirement today, he may be fired or laid off tomorrow. And he knows that this may happen in circumstances in which it will be impossible for him either to prove age discrimination and thus obtain compensation, because the employer can demonstrate the business necessity for his reduction in force, or to find alternative employment at an equivalent wage, especially if his current wage includes a return to firm-specific human capital. I have mentioned the plight of the elderly job-seeker before, and now I add that there is empirical evidence that older workers remain unemployed longer than younger ones in the wake of a plant closing or move, even after correction for the fact that they have a higher reservation wage as a result of having pension income to fall back on.27 It might actually pay an employer to engage in outright age discrimination from time to time to increase the incentive of older employees to elect early retirement. The demonstrated likelihood of such discrimination would increase the number of employ-ees who accepted the offer of early retirement, or would enable the employer to achieve his target number of acceptances with a less attractive early-retirement offer, or would do both. The rationally calculating employer would trade off this benefit of discrimi-nation against the cost in damages and other expenses of violating the Age Discrimi-nation in Employment Act. We have seen that these apparently are small. But of course they may be small because employers do not follow this Machiavellian strategy, and they may not do so because of the costs in the form of a bad reputation that an employer who treats his workers with such calculated ruthlessness would incur. He might have to pay for ruthlessness with higher wages.

Even if the employer is law abiding, and even if he does not announce a “Phase II” (involuntary termination), the employee who turns down an offer of early retirement takes a considerable risk. For if not enough employees accept the offer, the employer may decide to institute a RIF, though it was not his original intention to do so. Even if

26_{Public Employees Retirement System v. Betts, 492 U.S. 158 (1989). The provision of the age discrimination law on which}

Bettswas based, which exemptedbona fidebenefits plans, was modified by the Older Workers Benefit Protection Act, Pub. L. No. 101– 433, 104 Stat. 978 (1990) (codified at 29 U.S.C. § 621). But employers retain considerable latitude in withholding benefits to workers who turn down offers of early retirement. See Harper, note 24 above, at 1309 –1321.

27_{Douglas A. Love and William D. Torrence, “The Impact of Worker Age on Unemployment and Earnings after}

the RIF hits older and younger employees indiscriminately, some of the employees who refused offers of early retirement will be among those riffed, and they will be worse off than they would have been had they accepted the offer and worse off than younger workers, who will on average have better employment opportunities on the outside, would be.

Offers of early retirement are not a panacea from the standpoint of the employer concerned about the age distribution of his employees. He cannot be certain how many or which workers will accept the offer of early retirement. If he underestimates the acceptance rate, or if key workers elect early retirement, he may find himself having to incur the costs of rehiring workers to whom he has just paid early-retirement benefits.28 The “which” problem should be separated from the “how many” problem. Bountiful offers of early retirement might lead to an undesired reduction in the quality of the employer’s workforce. But this is unlikely. It is true that one group of workers likely to jump at such offers consists of those who have excellent alternative job prospects; any income from a new job would be in addition to their retirement benefits from the old one. These will tend to be the best workers—the most adaptable and energetic. The employer can try to hire them back— but his effort to do so will signal to them their value to him, and they will demand a high wage. However, this group of workers will be balanced by three other groups. The first consists of those who are particularly afraid that if they refuse the offer of early-retirement benefits, they will be terminated shortly anyway, without the benefits; and they will tend to be the poorer workers. The second group consists of those who attach a high value to leisure; they will tend to be the less committed and enthusiastic workers. The third group consists of those elderly workers who for reasons that may be unrelated to any particular fears of being fired or any unusual demand for leisure find early retirement a preferable alternative to continuing to work. These workers may be competent. But remember that, by hypothesis, the employer instituted the early-retirement program in an effort to reduce the average age of his workforce, so he derives a benefit from inducing a more or less random selection of elderly workers (random with regard to their competence) to leave.

What kind of worker brings an age discrimination suit? As a judge, I have been struck by the number of cases in which a salesman contends that he was fired because of his age. This impression is confirmed by my study. Of 271 cases for which the information is available, 25 (9.2%) were brought by salesmen, excluding retail sales clerks.29_{At first} glance, the large number of such cases is surprising. Salesmen are paid essentially on a piece-rate basis, so that if their productivity diminishes, whether because of age or for any other reason, their compensation and hence cost to the employer automatically falls.30_{But because there are fixed costs of employment, a reduction in wages will not}

28_{Frank E. Kuzmits and Lyle Sussman, “Early Retirement or Forced Resignation: Policy Issues for Downsizing}

Human Resources,”S.A.M. Advanced Management Journal, Winter 1988, pp. 28, 31; Mark S. Dichter and Mark A. Trank, “Learning to Manage Reductions-in-Force,”Management Review, March 1991, pp. 40, 41; William H. Honan, “New Law against Age Bias on Campus Clogs Academic Pipeline, Critics Say,”New York Times, June 15, 1994, p. B6; see also Joseph W. Ambash and Thomas Z. Reicher, “Proper Planning Key to Avoiding Discrimination Suits,”Pension World, May 1991, pp. 14, 15.

29_{William Graebner,A History of Retirement: The Meaning and Function of an American Institution, 1885–1978 44 – 49}

(New Haven: Yale University Press 1980), is a fascinating discussion of a wave of hostility to old salesmen in the period 1900 to 1924, when youthful physical energy was believed to be the key to successful salesmanship.

30_{Laurence J. Kotlikoff and Jagadeesh Gokhale, “Estimating a Firm’s Age-Productivity Profile Using the Present}

necessarily compensate the employer for the reduction in the worker’s output, just as in the parallel case of part-time work; so also if the salesman has an exclusive territory, so that if he falls down on the job, his employer’s sales will diminish.

The frequency of cases involving salesmen may be due to two factors. First, selling is more strenuous than most other white-collar work,31 _{and although this will induce} earlier retirement, it will also induce more discharges of workers who have not yet reached retirement age. Second, salesmen often have a large amount of firm-specific human capital (much of it social human capital—a network of valuable customer contacts). This makes it less likely that they will be fired, but if they are fired, it creates the prospect of very sizable damages, because their alternative wage is likely to be far below what they were receiving when they were fired. The largest award of damages in the sample was in fact obtained by a salesman. Employees whose principal human capital is general rather than specific so that their alternative wage will be close to the wage they were receiving when fired, or whose wages are in any event too small for the prospect of an award of lost wages to warrant the bother and expense of a lawsuit, will rarely show up as plaintiffs in age discrimination suits.

Consistent with this conjecture, most age discrimination suits are brought by profes-sional or managerial employees, who have high salaries; and most plaintiffs are in their fifties, have accumulated a lot of specific capital, and therefore have a substantial damage claim.32_{In my study, 165 of 271 cases were brought by professional, managerial,} or sales employees— 60.9%. This figure exceeds the share of these classes of workers in the working population as a whole. The largest number of plaintiffs in the sample were in the 55-to-59 age group, and the second largest number were in the 50-to-54 and 60-to-64 groups; only 22.8% were younger than 50. Young people would not have accumulated as much firm-specific human capital and, therefore, would have lower damages; elderly people (only 10.1% of the plaintiffs were 65 or older) would be either retired, and therefore no longer protected by a statute protecting employees (unless they had been involuntarily retired), or close to retirement, in which event their expected damages would be low, just as in the case of young workers.

As other studies have shown, blacks and women are under-represented as plaintiffs in age discrimination cases.33_{One explanation is that blacks and women do not “need” to} file an age discrimination suit as much as a white male, because they have other civil rights statutes to base a suit on. This is not a plausible explanation, because it is normally advantageous to sue on as many colorable claims as one has. A more plausible expla-nation is that blacks and women tend to have smaller investments in human capital, including firm-specific human capital, than white males.34_{The smaller that investment,} the less the expected gain from bringing an age discrimination suit.

31_{For evidence, see Pauline K. Robinson, “Age, Health, and Job Performance,” inAge, Health, and Employment63,}

692, (James E. Birren, Pauline K. Robinson, and Judy E. Livngston, eds., Englewood Cliffs, NJ: Prentice-Hall 1986).

32_{On both points, see Schuster and Miller, note 11 above, at 68 (Table 1); Schuster et al., note 11 above, at iii (59.3%}

of cases filed by managerial and professional employees). “The ADEA has become a grievance mechanism primarily utilized by white males and white-collar workers.” Christopher S. Miller et al., “State Enforcement of Age Discrimination in Employment Legislation” 18 (Syracuse University, School of Management, unpublished, n.d.).

33_{See id.; Rutherglen, note 12 above, at 510. I do not have the racial identity of the plaintiffs in my sample, but I}

do have their sex: 22% were women, which is only half their percentage of the total U.S. work force.

34_{For evidence regarding women, see, for example, John M. Barron, Dan A. Black, and Mark A. Loewenstein,}

“Gender Differences in Training, Capital, and Wages,” 28Journal of Human Resources343 (1994); Elizabeth Becker and Cotton M. Lindsay, “Sex Differences in Tenure Profiles: Effects of Shared Firm-Specific Investment,” 12Journal of Labor Economics98, 107–108 (1994); and Elisabeth M. Landes, “Sex-Differences in Wages and Employment: A Test of the

Given who the plaintiffs are, the Age Employment in Discrimination Act cannot realistically be characterized as progressive legislation. To the extent that the employer must factor into his labor costs the expected costs of damage judgments or settlements along with all the other costs of complying with (or violating) the Act, he will pay lower wages. He will try to lower the wages of those classes of workers most likely to bring and win age discrimination cases, but a perfect match-up cannot be expected, and this means that the costs of the Act will be borne in part at least by average workers. The Act’s effect is, thus, to redistribute wealth from younger to older workers, or, after it has been in effect for many years, from the same workers’ youth to their middle age; and, to a lesser extent, from average workers to members of the professional and managerial class (including commissioned salesmen), who account for most age discrimination cases.

Although redistribution of wealth from one’s young self to one’s old self is not the same thing as moving wealth from one’s left pocket to one’s right pocket, the welfare of the two selves is closely linked, in much the same way that the welfare of a man and of a woman is closely linked through joint consumption in marriage. So, efforts to prefer old over young, like efforts through sex discrimination law to prefer female over male, are unlikely to have substantial net redistributive effects. If sex-discrimination law increases employers’ labor costs, resulting in lower wages to male employees, the wives of those male employees will suffer.35_{Similarly, laws that redistribute wealth from young} to old harm the old to the extent that the old self internalizes the welfare of the young self.

Mandatory retirement. One might think that the 1986 amendment to the Age Discrim-ination in Employment Act that abolished mandatory retirement at any age in most occupations must have had big effects. But consider, first of all, that most workerswant

to retire when they reach the normal retirement age. A few must want to stay on; otherwise firms would never have imposed mandatory retirement. But many who do want to stay on can negotiate mutually satisfactory terms with employers for doing so.36 And if not, they can, as we have seen, be gently pried out, even after mandatory retirement has been abolished, by means of offers of early retirement on advantageous terms. It has been estimated that before the 1978 amendment to the Age Discrimination in Employment Act that raised the minimum mandatory retirement age from 65 to 70, only 5 to 10% of retired workers had been retired involuntarily.37 _{This may be an} overestimate. The year after the passage of the amendment saw no interruption in the steady downward trend in the percentage of persons 65 and older who were em-ployed,38_{although the downward trend in the labor-force participation rate of elderly}

Specific Capital Hypothesis,” 15Economic Inquiry523 (1977). I do not have evidence regarding firm-specific human capital of blacks.

35_{Richard A. Posner, “An Economic Analysis of Sex Discrimination Laws,” 56University of Chicago Law Review1311}

(1989). There is an offset: Husbands of wives who are benefited by laws against sex discrimination benefit also. But the offset is only partial, because fewer women than men work and working women are less likely to be married than working men.

36_{See generally Robert L. Kaufman and Seymour Spilerman, “The Age Structures of Occupations and Jobs,” 87}

American Journal of Sociology827 (1982).

37_{Philip L. Rones, “The Retirement Decision: A Question of Opportunity?”Monthly Labor Review, Nov. 1980, pp. 14,}

15; see also Joseph F. Quinn, Richard V. Burkhauser, and Daniel A. Myers,Passing the Torch: The Influence of Economic Incentives on Work and Retirement85– 87, 199 (Kalamazoo, MI: W.E. Upjohn Institute for Employment Research 1990).

men did finally bottom out in 1985, and it has risen moderately since. (The rise has been greater for women, but this means little; one would expect the labor-force participation rate of elderly women to be growing without regard to any legal changes, simply as a function of the greatly increased participation of women in the labor force in recent decades.) In addition to offering the carrot of early retirement, employers can wield a stick; they can require their workers to work harder, in the expectation that the disutility of working harder will fall disproportionately on the older workers and induce them to retire “voluntarily.”39_{It would be very difficult as a practical matter for a worker} to challenge such a tactic successfully under the age discrimination law.

Because education steepens the age-income profile—lower wages during the school-ing period beschool-ing made up by higher wages later— earnschool-ings peak later for the educated worker, and he is therefore more likely to want to work to an advanced age than a less educated worker. This implies that the abolition of mandatory retirement is likely to have benefited mainly the better educated, who are also likely to be the abler workers and therefore to have higher incomes beyond what is necessary merely to compensate them for their greater investment in human capital. This is further evidence that the Age Discrimination in Employment Act is regressive,40_{though only in the short run. It} is neutral in the long run because in the long run the highly educated will pay for the opportunity of working longer than their employer would like by accepting lower wages in their earlier years.

It is relevant to note that mandatory retirement did not become common until after World War II.41 _{Its advent coincided with the provision of pension benefits on a} systematic basis by large firms to all their long-time employees. The combination of a pension plan with mandatory retirement protects the worker against a penurious old age and the employer against having to pay a wage to a worker who is no longer productive yet may actually be receiving a wage premium to counteract his incentive to slack off in his last period.42 _{Even if the worker is not receiving an incentive wage} premium, and even if there is no contractual impediment to a reduction in his wage,43 an age-correlated decline in capability may have reduced his productivity to a point at which the employer would be better off replacing him. The employer could discharge the worker or force him to retire. But in the absence of mandatory retirement at a fixed age, the employer would have to make a costly and (to the worker) humiliating determination that the worker had ceased to be sufficiently productive to justify reten-tion. Because of its relation to incentive wage premia, union protections, and concern with avoiding stigma, mandatory retirement at fixed ages, rather than being a symptom

mandatory retirement age to 70, had no significant effect on the rate of participation in the labor force of persons aged 65 and over. Edward F. Lawlor, “The Impact of Age Discrimination Legislation on the Labor Force Participation of Aged Men: A Time-Series Analysis,” 10Evaluation Review794 (1986).

39_{Rones, note 37 above, at 15.}

40_{See Edward P. Lazear, “Why Is There Mandatory Retirement?” 87Journal of Political Economy1261, 1281–1283}

(1979).

41_{James H. Schulz,The Economics of Aging, ch. 3 (5th ed., NY: Auburn House 1992).}

42_{See Alan L. Gustman, Olivia S. Mitchell, and Thomas L. Steinmeier, “The Role of Pensions in the Labor Market:}

A Survey of the Literature,” 47Industrial and Labor Relations Review417, 431– 432 (1994), and studies cited there.

43_{And there may be, especially if the worker is covered by a collective bargaining agreement. Unions want employers}

to have as little discretion as possible to discharge workers, so it is no surprise that before mandatory retirement was abolished unionization and mandatory retirement were strongly positively correlated. Duane E. Leigh, “Why Is There Mandatory Retirement? An Empirical Reexamination,” 19Journal of Human Resources512, 525 (1984).

of the exploitation of older workers or the operation of a mindless ageism, is correlated with employment terms and practices that favor older workers.44

Mandatory retirement also may reduce an employer’s agency costs (the costs of aligning employees’ incentives with those of the employer). When involuntary retire-ment is discretionary, employees who want to work past the normal retireretire-ment age will have an incentive to forge alliances with their supervisors that may undermine the loyalty of both sets of employees, the supervised and the supervisory, to the employer. This concern may help explain why the federal civil service was one of the earliest institutions to adopt mandatory retirement. The goal was to sever the “personal ties and informal bonds” that continued to characterize relationships within the civil service long after the formal abolition of the spoils system.45

II. Sexual Harassment in the Workplace

I shall follow the same general format as in my discussion of age discrimination, examining first the underlying phenomenon and then the law’s response. But I will be briefer because there are fewer data.

The Causes, Incidence, and Character of Workplace Sexual Harassment

Sexual harassment in the modern workplace consists overwhelmingly of male harass-ment of females of reproductive age rather than male harassharass-ment of other men or of older women, or female harassment of males,46_{although these other types of} harass-ment are not unknown. Among the more common forms of male sexual harassharass-ment of young female employees are threats by male supervisors intended to extract sexual favors from female subordinates, sexual solicitation by male coworkers (including supervisors) of female workers, and verbal or other displays of sexual hostility or contempt by male workers toward female coworkers (for example, decorating the workplace with nude pinups). (In the last case, the victims need not be young.) I set to one side sexual harassment that involves forceful penetration or otherwise approaches rape; the offensiveness of such conduct is unproblematic. Not all peaceable sexual solicitations in the workplace, it should be noted, count as harassment. The law47_tries to distinguish between consensual dating, courtship behavior, and off-color banter, on the one hand, and seriously offensive solicitation on the other hand. Thus, a rule flatly forbidding dating between employees of the same employer, although it would be the most effective preventive measure against harassment, would impose a considerable cost by destroying an increasingly important part of the courtship market. With most women now entering the workforce before marriage, the workplace is an important site for

44_{As not only economists realize. See Carole Haber and Brian Gratton,Old Age and the Search for Security: An American}

Social History108 –109 (Bloomington, IN: Indiana University Press 1994). But see Martin Lyon Levine,Age Discrimination and the Mandatory Retirement Controversy(1988), a powerfully argued lawyer’s brief against mandatory retirement, questioning the economic rationale for the practice.

45_{Graebner, note 29 above, at 87.}

46_{See David M. Buss,The Evolution of Desire: Strategies of Human Mating160 (New York: Basic Books 1994); David E.}

Terpstra and Susan E. Cook, “Complainant Characteristics and Reported Behaviors and Consequences Associated with Formal Sexual Harassment Charges,” 38Personnel Psychology559, 562 (1985).

47_{Sexual harassment in the workplace is considered a form of sex discrimination. The principal law forbidding sex}

discrimination in employment, and so the principal vehicle for suing employers for sexual harassment, is Title VII of the Civil Rights Act of 1964, 42 U.S.C. § 2000eet seq.

courtship48_{and also a cheap one, because people can learn a lot about each other as a} byproduct of their ordinary work without having to date.

In the case of solicitations by a supervisor, the obvious explanation for why such an “offer” is resented is that it often carries with it an express or implied threat to fire or otherwise discriminate against the woman if she refuses. The resentment may be of the threat rather than of the solicitationper se. But the solicitation too may be resented as signaling the offeror’s unwillingness to recognize that the woman is of high rather than low status.49 _{A high-status woman, with good opportunities in the marriage market} (perhaps already married or having a high-status boyfriend), would have little to gain and much to lose from engaging in casual sex with her supervisor. A woman of low status would be more willing to consent to such a relationship. The solicitation is thus resented for much the same reason that a person might be offended to receive a cash gift from a friend. The making of such a gift would imply a belief that the person was hard up financially and therefore was of low status in our society, in which money is an important determinant of status.50_{It does not matter whether in fact a woman who is} solicited for casual sex is averse to it. Even if she is not, she may still be offended by the solicitation, because it implies that the solicitor thinks that she is of low status.

If the soliciting male has great power over the woman, the solicitation may not imply that he thinks she’s in the market for casual sex. He may simply think that although she does not want to have sex with him, she will do so to protect her job. Even in this case, however, there is an implication about her attitude toward casual sex—not that she likes it, but that she does not disvalue it as much as some other women would; she values her job more than her chastity. A related point is that because sexual solicitation in the workplace (as outside) often implies a violation of moral norms—the man or woman (or both) may be married, and the man may be promising the woman a competitive advantage over equally or better qualified coworkers—the solicitation signals the man’s belief that the woman is not a highly moral person.

The higher the man’s rank, however, the less likely the woman is to be offended.51 The man’s status may be so exalted that even a high-status woman would be flattered by his solicitation; in that event the solicitation would not imply a belief on his part that the woman was of low status and so she would have no reason to be offended. Moreover, there is always a possibility that the solicitation might lead to marriage with a high-status male or to other advantages.

Of course, not all sex is either casual or marital/reproductive. The male supervisor or coworker might be sincerely “smitten” by the object of his solicitations and might desire a passionately romantic rather than merely casually sexual relationship or a marital one. The range of possible meanings of a sexual solicitation complicates the interpretation

48_{See Carol Hymowitz and Ellen Joan Pollock, “Corporate Affairs: The One Clear Line in Interoffice Romance Has}

Become Blurred,”Wall Street Journal, Feb. 4, 1998, pp. A1, A8.

49_{Compare the finding that many men are willing to have casual sex with a woman they would not date, whereas}

many women are unwilling to have casual sex with some men whom they would date or with any men whom they would not date. See Douglas T. Kenrick et al., “Evolution, Traits, and the Stages of Human Courtship: Qualifying the Parental Investment Model,” 58Journal of Personality1 (1990); R. D. Clark and E. Hatfield, “Gender Differences in Receptivity to Sexual Offers,” 2Journal of Psychology and Human Sexuality39 (1989).

50_{One would therefore expect cash gifts to be more common in societies in which money is a less important}

determinant of status than it is in our society. In addition, cash gifts are presumably more common within families, inasmuch as status within a family is not determined primarily by money.

employer of detecting the harassment may exceed the benefits to him. In that event there will be no remedial action even if the total costs of the harassment exceed the total benefits. And in such a case the economic rationale for punishing the harasser might seem to be similar to the economic rationale for punishing embezzlement, which also occurs as an incident to a voluntary employment relationship and so might also be thought to involve no externality. But it does. If the expected costs of punishment, when added to any other costs borne by the embezzler, are less than the costs that the embezzlement inflicts on society, the embezzler will have succeeded in externalizing some of the costs of his misconduct. This is the economic justification for subjecting the embezzler to criminal punishment, rather than to the lesser “punishment” (discharge, a bad employment reference, a civil suit for conversion, etc.) available to the employer. The law punishes the embezzler rather than the employer because the embezzler is the wrongdoer. But it punishes the employer rather than the harasser even though the latter is the wrongdoer. The reasons for the difference in treatment may be that harassment is not a serious enough wrong to warrant the costs of criminal punishment, that it is difficult to define with the precision desirable in criminal statutes, and that although the harasser would in principle be liable in a civil tort suit to the person harassed (just as the embezzler would in principle be liable in a civil tort suit, for conversion, by his employer), he will often be unable to pay a tort judgment and will therefore not be worth suing. If for these reasons the harasser is, as a practical matter, liable to neither criminal nor civil sanctions, then by making the employer liable for the harm done by the harassing employee the law enlists the employer as an indispensable supplementary law enforcer. The method resembles the doctrine ofrespondeat superiorin tort law, which makes an employer liable for the torts committed by his employees within the scope of the tortfeasor’s employment. The economic rationale of the doc-trine, if the victim of the employee’s tort is a stranger to the employer’s enterprise (for example, a pedestrian run over by a truck driven by an employee of the truck’s owner) rather than another employee,58_{and if the tortfeasing employee is judgment-proof, is} that the cost of the tort will be externalized unless the employer is liable. Liability will give him an incentive to take cost-justified measures in selecting and monitoring his employees to reduce the likelihood of their committing torts for which he will have to pay.59

But a law that makes the employer liable for sexual harassment by his employees cannot be defended on the ground just sketched, because the imposition of liability for workplace sexual harassment does not correct an externality. Theharasserimposes an externality to the extent that the expected cost of punishment to him is less than the social cost of the harassment; and that externality could be internalized by criminal punishment. But it is an externality to the harasser, not to the employer, because the employer has a contractual relation with the victim of the harasser, another employee. The Coase theorem clicks in and implies that, at least as a first approximation, the

58_{Injuries of one employee by another employee of the same employer are nowadays in American law mostly}

governed by workers’ compensation statutes rather than by tort law. The tort law of workplace injuries contains a doctrine, the “fellow-servant rule,” that has, as we shall see, been carried into the law of sexual harassment to prevent an employer from being held liable for coworker harassment unless he knows or should know about the harassment and fails to take cost-justified measures against it. On the economizing properties of the fellow-servant rule, see William M. Landes and Richard A. Posner,The Economic Structure of Tort Law309 –310 (Cambridge, MA: Harvard University Press 1987).

59_{See Posner, note 14 above, § 6.8, pp. 204 –206; Alan O. Sykes, “The Boundaries of Vicarious Liability: An Economic}

imposition of liability on the employer changes only the form in which female workers are compensated for being harassed—instead of receiving a wage that compensates themex antefor the risk of being harassed, as they would if employers were not liable for sexual harassment, they receive a lower wage coupled with a tort entitlement toex postdamages if they are harassed. This implies that if women were free to waive their rights under Title VII to sue for sexual harassment, they would sell them to employers for a higher wage. For if ex post compensation were the most efficient method of compensating for this workplace disamenity, employers would probably adopt it without the prodding of the law. It probably is not efficient, given the cost of litigation, which is why both employer and employee might well be better off by an agreement whereby the latter would waive her right to sue in exchange for higher pay or other benefits.

The picture is complicated by the fact that advance waivers of Title VII rights are forbidden60_{and that employers are forbidden by the Equal Pay Act}61_{to pay men and} women different wages for the same work. As a result, part of an employer’s higher costs from being subject to sexual harassment law are borne by male coworkers. The essential point, however, is that employers have an incentive wholly apart from the law to take measures against sexual harassment, because employers who fail to do so will have to compensate their female employees in higher wages. The incentive may be strong enough to optimize the degree of harassment to which female workers are exposed, at least in private-sector employment, where incentives to minimize costs are strongest.

Laws against sexual harassment have the interesting side-effect of altering the char-acterof such harassment in such a way that the public may mistakenly believe both that there ismoreharassment than ever (because there will be more complaints if there are more remedies, although an offsetting factor is the deterrent effect of the remedies on the incidence of conduct giving rise to complaints) and that harassment takes grosser forms than it would were it not for the law. Before there is a law against harassment, the distribution of men who practice it will be similar to the distribution of men generally, except that the most refined, religious, or sexually inert men will not be represented. After the law goes into effect, the distribution alters; the law-abiding, defined as those who are more responsive to legal sanctions, whether because they have superior “char-acter,” better sexual alternatives, or a substantial reputation or status capital that will be impaired by the imposition of such sanctions, fall out. The men who remain are a now a less representative, and more unsavory, segment of the total male distribution.62 Sexual perverts, psychopaths, and other disordered types will now constitute a larger fraction of the harassers, along with men who are insensitive, coarse, or stupid.

The general point is that by altering the distribution of actors in an activity, the law can make the activity seem more unsavory, more threatening, than it would be if it were not outlawed.63_{Through this selection effect, the making of an activity unlawful may}

60_{That is, a waiver signed in advance of the harassment. Once the harassment has occurred, and the employee thus}

has a legal claim, she can waive it in exchange for compensation, just as any legal claim can be settled. SeeAlexander v. Gardner-Denver Co., 415 U.S. 36 (1974).

61_{29 U.S.C. § 206(d).}

62_{John R. Lott, Jr., “Do We Punish High Income Criminals Too Heavily?” 30Economic Inquiry583 (1992), argues that}

reputation effects are likely to be greater than the direct disutility of sanctions when the sanctions are directed against upper-income people.

63_{This is a pronounced effect of the law against trafficking in drugs, which has had the effect of associating the sale}

of drugs with social outcasts. If the drug laws were repealed, drugs would be sold by respectable people, for example pharmacists. Cf. Richard A. Posner, “Social Norms, Social Meaning, and Economic Analysis of Law: A Comment,” 27 Journal of Legal Studies553, 562 (1998).

strengthen the (apparent) case for making it unlawful. The effect, however, is likely to be less pronounced in the case of harassment than in the case of drug dealing, because one response to threatened punishment for harassment is to engage in milder forms of harassment, inasmuch as they are less likely to give rise to sanctions.

The Political Economy of Sexual Harassment Laws

Just as in the case of the age discrimination law, a law forbidding sexual harassment may not on balance benefit the protected group. It may make employers more reluctant to hire women in jobs in which sexual harassment is likely, although a possible offsetting effect is to make employers more reluctant to hire men. Consider a work force that consists of all women, and now a man applies for a vacancy. The employer might refuse to hire him, fearing that to do so would subject the employer for the first time to potential liability for sexual harassment. But, conversely, the employer whose work force is all male will for the same reason be reluctant to hire a woman. In mixed work forces, the effect of a sexual harassment law on the relative propensity to hire men or women will depend on such things as the propensity of women to complain about real and imagined harassment and the cost of screening prospective employees for potential female complainers and potential male harassers.

Another effect of the law is to give employers an incentive to segregate their work force by sex (as by steering women away from jobs in which they would be traveling with male employees) or, more broadly, to reduce women’s workplace courtship opportu-nities, which as we noted earlier are increasingly important. The law may also place employers on a razor’s edge where they are liable for sexual harassment if they do not maintain a close surveillance of their workers and are liable for invasion of privacy if they do. Finally, by increasing an employer’s labor costs, and hence by reducing wages generally, sexual harassment law may harm married women employees because their welfare tends to decline when their husbands’ incomes decline, because much house-hold consumption is joint.

If the average woman and the average man are both harmed by sexual harassment laws, the question arises as to why we have these laws. The answer may be that although women as a whole lose, some women and men gain, and they may be more effective politically than the losers. The clearest gainers are women who are and expect to remain single, because their welfare is not tied up with that of men. If the employer for practical or legal reasons cannot discriminate in wages between his male and female employees, then the male employees will bear a part of any higher labor costs that are due to women’s rights to complain about workplace sexual harassment. Married nonworking women will not benefit from the transfer, and this may be part of the explanation for why the sexual harassment tort is of recent origin (the 1980s). When most women worked at home, and those who worked outside generally had low-level jobs, the demand for protection against sexual harassment would have been slight. The demand grew as more and more women entered the workforce until finally there was a critical mass able to obtain a change in the law.

The Form of the Law

In the case of age discrimination, the issue of discrimination and the issue of the employer’s liability for the discrimination tend to merge. If the plaintiff can show that the employer discriminated against him on account of his age, he is entitled to a judgment, even though the actual discriminator will be another employee—the

super-visory employee who fired, demoted, refused to hire, or otherwise harmed the plaintiff. The doctrine ofrespondeat superiormakes the employer liable for torts committed by his employees within the scope of their employment. But in the usual case of sexual harassment, the conduct is not within the scope of the harasser’s employment. That is, it is not a consequence of the harasser’s trying to carry out, however ineptly or even maliciously, the job responsibilities that he has been given by the employer. He is engaged in a “frolic” of his own for the consequences of which an employer normally would not be liable unless the employer had been negligent in the selection, training, or supervision of the employee.

As clarified recently by a pair of Supreme Court decisions, the scope of employer liability in sexual harassment cases is approximately as follows.64 _{If the harasser was} indeed engaged in a frolic of his own and thus was not acting within the scope of his employment, the employer will be liable only if the employer was negligent in the selection, training, or supervision of the employee—provided that the harasser was a coworker65_{of the victimized employee rather than a supervisor. If the victimizer was a} supervisor, the employer is strictly liable if the harassment took the form of a “company act” (I will explain that term shortly). If, however, the victimizer was a supervisor but the harassment did not take the form of a company act, the employer is only prima facie

strictly liable; for he has a complete defense to liability if he can prove, first, that he had a good internal procedure for responding to harassment complaints promptly and with appropriate and effective remedial action, and, second, that the victim failed to use the procedure.

This pattern66_{makes a fair amount of economic sense, though it would be simpler to} make the employer strictly liable for company acts and liable only if negligent in all other cases. The difference between this and the more complex pattern that the Supreme Court has imposed has primarily to do with which party to the suit has the burden of proof on the issue of negligence, and that is likely to be important only in very close cases.

Begin with coworker harassment. The argument against strict liability is that the employer could not stamp out this sort of harassment without going to extreme expense and greatly curtailing the privacy of its employees, as by putting them under continuous video surveillance. If the victim of sexual harassment complains to a supervisor, or, if a worker who notices what is going on complains to a supervisor, or if the harassment is so pervasive (considering its nature, frequency, and number of victims and perpetra-tors) that the employer knows or should know about the harassment, the employer ought to take steps to correct the problem. He has or should have the information; he has only to act on it. And because everyone knows by now that sexual harassment is a common problem in the American workplace, the employer ought in addition to take, in advance of specific cases of harassment, preventive measures against it, as by adopting and announcing a policy against sexual harassment and creating a discreet and conve-nient machinery by means of which victims can obtain relief without exposing them-selves to retaliation. These are the responsibilities that a negligence standard imposes. A standard of liability that asks the employer to do more than is feasible to control harassment will impose costs without creating deterrent benefits. If the law imposes

64_{Burlington Industries, Inc. v. Ellerth, 118 S. Ct. 2257 (1998);Faragher v. City of Boca Raton, 118 S. Ct. 2275 (1998).} 65_{This is the “fellow servant” component of sexual harassment law. See note 58 above.}

66_{Imposed as an interpretation of Title VII, therefore not necessarily applicable to sexual harassment case brought}

strict liability, so that the employer is liable even when there is no reasonable measure he could have taken to prevent the harassment, the only effect will be to impose extra costs on employers and those with whom they are linked contractually (including, as we have emphasized, their employees). Employers will prefer paying the occasional judg-ment to incurring costs that, by definition, exceed the employer’s foreseeable liabili-ty— by definition because, were the costs less than the expected liability, the failure to incur them would be negligence and so strict liability would make no difference. Strict liability bites only when the cost of prevention exceeds the expected cost of a legal judgment.67

Consider now cases in which the harasser is a supervisor rather than a line employee. We must distinguish between two types of sexual harassment by supervisors. In the first type, the supervisor uses or attempts to use his supervisory authority to obtain sexual favors from an employee. This is the domain of what is called “quid pro quo” harassment. In the second type of case, the supervisor does not use or attempt to use his supervisory authority at all. He harasses an employee, but he does so in exactly the same way that an employee who had no supervisory authority would harass another employee. The proper standard of employer liability here is negligence, just as in the case of harass-ment by nonsupervisory employees, because it is as costly for the employer to police this kind of harassment by a supervisor as it is to police the identical harassment by a coworker. The employer may have thousands of supervisory employees, and because they will endeavor to keep any harassment activity in which they engage private, it will often be impossible for the employer, without excessively intrusive monitoring, to detect supervisory harassment. If the employer is small and has few supervisory employees, failure to detect harassment may show that it has been negligent in its supervision of its supervisors.

Turning now toquid pro quoharassment, I again make a twofold distinction. There is first the case in which the supervisor brings about a significant alteration in the terms or conditions of his victim’s employment. He fires her, or denies her a promotion, or blocks a scheduled raise, or demotes her, or transfers her to a less desirable job location, or refuses to give her the training that the company’s rules entitle her to receive. In all these examples the supervisor is using his delegated authority to perform a company act. Strict liability is appropriate in this kind of case because it is likely to deter this kind of sexual harassment much more effectively than negligence liability would, and at a reasonable cost. The employer who is strictly liable will monitor the exercise of this delegated authority very carefully, knowing that he will be liable if the authority is abused. This monitoring should be relatively easy to do, because it is usually a mistake for a firm, quite without regard to any potential legal liability, to give a supervisor unilateral authority to alter a subordinate’s terms or conditions of employment signif-icantly. In well-managed companies, decisions having such consequences are subject to rules and to review by higher-ups in the company—the industrial equivalent of appel-late review. The rules will be more carefully formuappel-lated and the supervisor’s compliance with them in firing or otherwise hurting a subordinate more carefully reviewed by the supervisor’s superiors if the employer is strictly liable for the supervisor’s use of dele-gated powers to harass subordinates. Courts applying a negligence standard would have great difficulty determining how closely the supervisor should be supervised. Such

67_{This is a somewhat oversimplified analysis of the economic difference between negligence and strict liability, but}

questions as how many tiers of review should be provided before an employee can be fired or demoted are not easily answered in terms of reasonableness or due care, the criteria of negligence. The regime of strict liability shifts the responsibility for deciding these questions to the employer, who knows more than a court does about how to control their supervisory employees.

In contrast, courts do know, more or less anyway, what is reasonable for an employer to do about hostile-environment harassment—institute a tough policy, disseminate it, establish a procedure by which a worker can complain without fear of retaliation, and respond promptly and effectively to any report of possible harassment. Knowing what the employer should do, the courts have only to decide whether he did it to decide whether he was negligent and should be liable. When it comes to designing the optimum system for reigning in the discretion of supervisory employees, however, the courts are at sea.

Strict liability is inappropriate, however, when the supervisor merely makes threats, even if the threats are effective. That is why it is important to distinguish between the case in which the supervisor actually alters the terms or conditions of his victim’s employment (that is, he commits a company act) and the case in which he merely threatens to do so, whether or not the victim yields to the threats. Suppose the supervisor threatens to fire a subordinate unless she will have sex with him and she agrees— or she refuses and he does not carry out his threat. In either case, because the supervisor has not used his delegated authority to commit a company act, a system for vetting such acts would not detect his misconduct. In a well-regulated company a supervisor who wants to fire a subordinate has to obtain the approval of higher-ups, as we have said, and so they will have an opportunity to determine the bona fides of his proposal. But if he does not propose to fire the subordinate, whether because she has submitted to his sexual extortion or called his bluff, there will be no proposed action for higher-ups to review. It will be no more feasible for the company to determine what is going on than it would be if the harasser were a coworker who had threatened to steal the victim’s work tools if she did not submit to him.

I conclude that the current structure of the sexual harassment tort when litigated under Title VII makes economic sense. That leaves open, however, the fundamental question, on which I have expressed skepticism, of whether the tort itself makes economic sense.